Predicting expansion failures and defragmenting cluster resources

Abstract

The present disclosure relates to systems, methods, and computer readable media for predicting expansion failures and implementing defragmentation instructions based on the predicted expansion failures and other signals. For example, systems disclosed herein may apply a failure prediction model to determine an expansion failure prediction associated with an estimated likelihood that deployment failures will occur on a node cluster. The systems disclosed herein may further generate defragmentation instructions indicating a severity level that a defragmentation engine may execute on a cluster level to prevent expansion failures while minimizing negative customer impacts. By uniquely generating defragmentation instructions for each node cluster, a cloud computing system can minimize expansion failures, increase resource capacity, reduce costs, and provide access to reliable services to customers.

What is claimed is: 
     
       1. A method, comprising:
 identifying a plurality of cluster features for a node cluster based on utilization data for the node cluster, wherein the node cluster comprises a grouping of server nodes on a cloud computing system; 
 determining one or more expansion failure metrics for the node cluster based on the identified plurality of cluster features, at least one of the expansion failure metrics including an expansion failure prediction for a set of deployments on the node cluster; 
 generating defragmentation instructions applicable to the node cluster based on the one or more expansion failure metrics, the defragmentation instructions indicating a defragmentation severity level for performing one or more defragmentation actions on the node cluster; and 
 providing the defragmentation instructions to a server device associated with the node cluster, wherein providing the defragmentation instructions to the server device causes the server device to perform the one or more defragmentation actions on server nodes of the node cluster. 
 
     
     
       2. The method of  claim 1 , wherein the cluster features include one or more of:
 a current availability of empty server nodes on the node cluster; 
 a difference between a current core utilization on the node cluster and a total capacity of compute cores on the node cluster; or 
 a fragmentation metric based on a ratio of available compute cores in a set of server nodes and a total number of compute cores on the set of server nodes. 
 
     
     
       3. The method of  claim 2 , wherein the cluster features include one or more of:
 property log signals including characteristics associated with hardware types of the node cluster and virtual machine types deployed on the node cluster; or 
 event log signals including update request information associated with the set of deployments on the node cluster. 
 
     
     
       4. The method of  claim 1 , wherein the expansion failure prediction includes a risk classification for the node cluster, the risk classification indicating an estimated probability that the set of deployments on the node cluster will experience one or more expansion failures within a threshold period of time. 
     
     
       5. The method of  claim 4 , wherein the risk classification includes a risk category from a plurality of possible risk categories, each of the risk categories corresponding to a different defragmentation severity level for performing the one or more defragmentation actions on the node cluster. 
     
     
       6. The method of  claim 1 , wherein the one or more expansion failure metrics further includes one or more of:
 an indication of one or more expansion failures that have occurred within a threshold period of time; or 
 a current availability of empty server nodes on the node cluster. 
 
     
     
       7. The method of  claim 1 , wherein the one or more expansion failure metrics includes one or more cluster-specific fragmentation parameters associated with a threshold performance level for the node cluster, the one or more cluster-specific fragmentation parameters including one or more of:
 a minimum availability of empty server nodes on the node cluster; or 
 a target availability of empty server nodes on the node cluster. 
 
     
     
       8. The method of  claim 7 , further comprising:
 receiving a user input identifying the one or more cluster-specific fragmentation parameters, 
 wherein generating the defragmentation instructions comprises determining the defragmentation severity level based on a combination of the expansion failure prediction and the received user input identifying the one or more cluster-specific fragmentation parameters. 
 
     
     
       9. The method of  claim 1 , further comprising:
 identifying additional cluster features for a second node cluster based on additional utilization data for the second node cluster; 
 determining additional expansion failure metrics for the second node cluster based on the identified additional cluster features; and 
 generating additional defragmentation instructions applicable to the second node cluster based on the additional expansion failure metrics. 
 
     
     
       10. The method of  claim 9 , wherein the defragmentation instructions and the additional defragmentation instructions each include an indicated level of priority for performing defragmentation on the node cluster and the second node cluster, and wherein the method further comprises determining an order of implementing defragmentation actions on the node cluster and the second node cluster based on the indicated level of priority for each of the node cluster and the second node cluster. 
     
     
       11. The method of  claim 1 , wherein the one or more defragmentation actions includes live migrating one or more virtual machines between server nodes to defragment the current computing capacity on the node cluster. 
     
     
       12. The method of  claim 1 , wherein the one or more defragmentation actions includes live migrating one or more virtual machines to consolidate workloads on fragmented nodes to increase a number of empty nodes on the node cluster. 
     
     
       13. A system, comprising:
 one or more processors; 
 memory in communication with the one or more processors; 
 instructions stored in the memory, the instructions being executable by the one or more processors to:
 identify a plurality of cluster features for a node cluster based on utilization data for the node cluster, wherein the node cluster comprises a grouping of server nodes on a cloud computing system; 
 determine one or more expansion failure metrics for the node cluster based on the identified plurality of cluster features, at least one of the expansion failure metrics including an expansion failure prediction for a set of deployments on the node cluster; 
 generate defragmentation instructions applicable to the node cluster based on the one or more expansion failure metrics, the defragmentation instructions indicating a defragmentation severity level for performing one or more defragmentation actions on the node cluster; and 
 provide the defragmentation instructions to a server device associated with the node cluster, wherein providing the defragmentation instructions to the server device causes the server device to perform the one or more defragmentation actions on server nodes of the node cluster. 
 
 
     
     
       14. The system of  claim 13 , wherein the cluster features include one or more of:
 a current availability of empty server nodes on the node cluster; 
 a difference between a current core utilization on the node cluster and a total capacity of compute cores on the node cluster; 
 a fragmentation metric based on a ratio of available compute cores in a set of server nodes and a total number of compute cores on the set of server nodes; 
 property log signals including characteristics associated with hardware types of the node cluster and virtual machine types deployed on the node cluster; or 
 event log signals including update request information associated with the set of deployments on the node cluster. 
 
     
     
       15. The system of  claim 13 ,
 wherein the expansion failure prediction includes a risk classification for the node cluster, the risk classification indicating an estimated probability that the set of deployments on the node cluster will experience one or more expansion failures within a threshold period of time, and 
 wherein the risk classification includes a risk category from a plurality of possible risk categories, each of the risk categories corresponding to a different defragmentation severity level for performing the one or more defragmentation actions on the node cluster. 
 
     
     
       16. The system of  claim 13 , wherein the one or more expansion failure metrics further includes one or more of:
 an indication of one or more expansion failures that have occurred within a threshold period of time; or 
 a current availability of empty server nodes on the node cluster. 
 
     
     
       17. The system of  claim 13 , wherein the one or more failure metrics includes one or more cluster-specific fragmentation parameters including one or more of a minimum availability of empty server nodes on the node cluster or a target availability of empty server nodes on the node cluster, and further comprising instructions being executable by the one or more processors to:
 receive a user input identifying the one or more cluster-specific fragmentation parameters, wherein generating the defragmentation instructions comprises determining the defragmentation severity level based on a combination of the expansion failure prediction and the received user input identifying the one or more cluster-specific fragmentation parameters. 
 
     
     
       18. A non-transitory computer readable medium storing instructions thereon that, when executed by one or more processors, causes a computing device to:
 identify a plurality of cluster features for a node cluster based on utilization data for the node cluster, wherein the node cluster comprises a grouping of server nodes on a cloud computing system; 
 determine one or more expansion failure metrics for the node cluster based on the identified plurality of cluster features, at least one of the expansion failure metrics including an expansion failure prediction for a set of deployments on the node cluster; 
 generate defragmentation instructions applicable to the node cluster based on the one or more expansion failure metrics, the defragmentation instructions indicating a defragmentation severity level for performing one or more defragmentation actions on the node cluster; and 
 provide the defragmentation instructions to a server device associated with the node cluster, wherein providing the defragmentation instructions to the server device causes the server device to perform the one or more defragmentation actions on server nodes of the node cluster. 
 
     
     
       19. The non-transitory computer readable medium of  claim 18 , wherein the cluster features include one or more of:
 a current availability of empty server nodes on the node cluster; 
 a difference between a current core utilization on the node cluster and a total capacity of compute cores on the node cluster; 
 a fragmentation metric based on a ratio of available compute cores in a set of server nodes and a total number of compute cores on the set of server nodes; 
 property log signals including characteristics associated with hardware types of the node cluster and virtual machine types deployed on the node cluster; or 
 event log signals including update request information associated with the set of deployments on the node cluster. 
 
     
     
       20. The non-transitory computer readable medium of  claim 18 , wherein the one or more failure metrics includes one or more cluster-specific fragmentation parameters including one or more of a minimum availability of empty server nodes on the node cluster or a target availability of empty server nodes on the node cluster, wherein the instructions, when executed by the one or more processors, causes the computing device to:
 receive a user input identifying the one or more cluster-specific fragmentation parameters, wherein generating the defragmentation instructions comprises determining the defragmentation severity level based on a combination of the expansion failure prediction and the received user input identifying the one or more cluster-specific fragmentation parameters.